This study examines how AI-powered conversational agents (CAs) can assist older adults in learning and interacting with virtual reality (VR) technology. Using semi-structured interviews and a participatory design workshop, we investigated the challenges older adults encounter when learning VR, their support needs, and their expectations from AI assistance. The results highlight physical obstacles, complex interfaces, and social-contextual issues in VR learning. Older adults expressed a preference for personalized, adaptable support that preserves their independence. They anticipated AI-driven CAs to enable natural interactions, offer context-sensitive help, and maintain empathetic communication. Drawing from these insights, we offer design recommendations for AI-powered CAs that strike a balance between functionality and ease of use, incorporate emotional intelligence, and adapt to various social and environmental settings. This research contributes to the development of age-friendly technology and provides practical guidance for creating AI-enhanced VR learning systems that encourage lifelong engagement.
